platform/upstream/gstreamer.git
4 years agogroup: fix priority setting
Henry Wilkes [Sat, 18 Apr 2020 15:49:31 +0000 (16:49 +0100)]
group: fix priority setting

Stop moving the group if a child clip is being edited by timeline-tree,
a child group is updating its own priority, or a layer that a clip is in
has changed priority. A group should only move if a descendant moves
layers outside of a timeline-tree edit, or the priority of the group is
set by the user.

Fixes https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/issues/89

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/169>

4 years agocontainer: keep start and duration up to date
Henry Wilkes [Sat, 18 Apr 2020 15:34:56 +0000 (16:34 +0100)]
container: keep start and duration up to date

Simplified keeping the start and the duration of a container/group up to
date with the earliest start of the children and the last end of the
children. The previous logic was spread between ges-group and
ges-container, now all the position handling is in ges-container.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/169>

4 years agouri-clip: use duration-limit in set_max_duration
Henry Wilkes [Tue, 28 Apr 2020 17:01:04 +0000 (18:01 +0100)]
uri-clip: use duration-limit in set_max_duration

Use the duration-limit rather than max-duration - in-point, since the
former will be able to take other factors, such as effects, into
account.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/169>

4 years agoclip: add the duration-limit property
Henry Wilkes [Mon, 13 Apr 2020 16:42:22 +0000 (17:42 +0100)]
clip: add the duration-limit property

The duration-limit is the maximum duration that can be set for the clip
given its current children and their properties. If a change in the
children properties causes this to drop below the current duration, it
is automatically capped by this limit.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/169>

4 years agoges: Output otio formatter loading issues in debug logs
Thibault Saunier [Mon, 4 May 2020 14:35:25 +0000 (10:35 -0400)]
ges: Output otio formatter loading issues in debug logs

Instead of spamming the terminal with a python traceback

Fixes https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/issues/107

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/173>

4 years agotests: Stop recording segment position in seek_with_stop
Thibault Saunier [Wed, 6 May 2020 03:03:36 +0000 (23:03 -0400)]
tests: Stop recording segment position in seek_with_stop

There are two valid timing in GstAggregator where the segment event
is pushed before GstAggregator sets its srcpad->segment.position in
gst_aggregator_pad_chain_internal. Segment.position is basically
a helper field for internal elements use so we should not require
a specific value here as we are not checking a particular element
behavior.

Fixes https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/issues/106

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/174>

4 years agotest: Add support for .validatetest in the launcher app
Thibault Saunier [Sat, 2 May 2020 05:24:18 +0000 (01:24 -0400)]
test: Add support for .validatetest in the launcher app

4 years agocontainer: return TRUE if adding doesn't cause any errors
Henry Wilkes [Fri, 1 May 2020 13:26:32 +0000 (14:26 +0100)]
container: return TRUE if adding doesn't cause any errors

If `add_child` and `set_parent` succeed we want to return TRUE, even if
the added element is no longer a child by the end of the method. This is
because some users may call ges_container_remove during `child-added`.
This shouldn't be considered an error.

4 years agoges:tests: Fix the `ignore-fields` format in validatetests
Thibault Saunier [Thu, 30 Apr 2020 21:44:33 +0000 (17:44 -0400)]
ges:tests: Fix the `ignore-fields` format in validatetests

They are needed as those are not 100% reproducible with GES.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/168>

4 years agoplugin: Fix a race removing tracks from timeline from the wrong thread
Thibault Saunier [Thu, 30 Apr 2020 17:23:05 +0000 (13:23 -0400)]
plugin: Fix a race removing tracks from timeline from the wrong thread

The case was that the timeline state was being changed from the parent
composition's action thread before the timeline was committed, leading
to the SELECT_STREAM event to be pushed from the track to the nested
timeline from the wrong composition thread.

```
**
GES:ERROR:../subprojects/gst-editing-services/ges/ges-track.c:1263:ges_track_remove_element: assertion failed: (track->priv->valid_thread == g_thread_self())
Bail out! GES:ERROR:../subprojects/gst-editing-services/ges/ges-track.c:1263:ges_track_remove_element: assertion failed: (track->priv->valid_thread == g_thread_self())

   Thread 1 (Thread 0x7f6ec2d43700 (LWP 1228982)):
   #0  0x00007f6ed85b2a25 in raise () at /lib64/libc.so.6
   #1  0x00007f6ed859b895 in abort () at /lib64/libc.so.6
   #2  0x00007f6ed899cb8c in g_assertion_message (domain=<optimized out>, file=0x7f6ed8d7fd58 "../subprojects/gst-editing-services/ges/ges-track.c", line=<optimized out>, func=<optimized out>, message=<optimized out>) at ../glib/gtestutils.c:2914
   #3  0x00007f6ed89fa9ff in g_assertion_message_expr (domain=domain@entry=0x7f6ed8d76875 "GES", file=file@entry=0x7f6ed8d7fd58 "../subprojects/gst-editing-services/ges/ges-track.c", line=line@entry=1263, func=func@entry=0x7f6ed8d805b0 <__func__.6> "ges_track_remove_element", expr=expr@entry=0x7f6ed8d801e8 "track->priv->valid_thread == g_thread_self()") at ../glib/gtestutils.c:2940
   #4  0x00007f6ed8d2658f in ges_track_remove_element (track=track@entry=0x7f6eb4119b20 [GESAudioTrack], object=object@entry=0x106f240 [GESAudioUriSource]) at ../subprojects/gst-editing-services/ges/ges-track.c:1263
   #5  0x00007f6ed8d10842 in ges_clip_empty_from_track (clip=0x7f6e7803ee80 [GESUriClip], track=track@entry=0x7f6eb4119b20 [GESAudioTrack]) at ../subprojects/gst-editing-services/ges/ges-clip.c:1086
   #6  0x00007f6ed8d01453 in ges_timeline_remove_track (timeline=timeline@entry=0x7f6e6c01ae50 [GESTimeline], track=0x7f6eb4119b20 [GESAudioTrack]) at ../subprojects/gst-editing-services/ges/ges-timeline.c:2460
   #7  0x00007f6ed8d0286b in ges_timeline_send_event (element=<optimized out>, event=<optimized out>) at ../subprojects/gst-editing-services/ges/ges-timeline.c:484
   #8  0x00007f6ed8bf466c in gst_element_send_event (element=0x7f6e6c01ae50 [GESTimeline], event=event@entry=0x7f6eb410f9f0) at ../subprojects/gstreamer/gst/gstelement.c:1934
   #9  0x00007f6ed8d242cd in ges_track_handle_message (bin=0xd846f0 [GESVideoTrack], message=0x7f6eb411ac90) at ../subprojects/gst-editing-services/ges/ges-track.c:477
   #10 0x00007f6ed8bc9128 in bin_bus_handler (bus=<optimized out>, message=<optimized out>, bin=<optimized out>) at ../subprojects/gstreamer/gst/gstbin.c:3286
   #11 0x00007f6ed8bdbae2 in gst_bus_post (bus=bus@entry=0xfdf440 [GstBus], message=message@entry=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstbus.c:359
   #12 0x00007f6ed8bf1396 in gst_element_post_message_default (element=element@entry=0x10261d0 [NleComposition], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2067
   #13 0x00007f6ed8bccbee in gst_bin_post_message (element=0x10261d0 [NleComposition], msg=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstmessage.h:376
   #14 0x00007f6ed8bf4b66 in gst_element_post_message (element=0x10261d0 [NleComposition], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2110
   #15 0x00007f6ed8bc9128 in bin_bus_handler (bus=<optimized out>, message=<optimized out>, bin=<optimized out>) at ../subprojects/gstreamer/gst/gstbin.c:3286
   #16 0x00007f6ed8bdbae2 in gst_bus_post (bus=bus@entry=0xfdf2c0 [GstBus], message=message@entry=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstbus.c:359
   #17 0x00007f6ed8bf1396 in gst_element_post_message_default (element=element@entry=0x1029110 [GstBin], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2067
   #18 0x00007f6ed8bccbee in gst_bin_post_message (element=0x1029110 [GstBin], msg=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstmessage.h:376
   #19 0x00007f6ed8bf4b66 in gst_element_post_message (element=0x1029110 [GstBin], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2110
   #20 0x00007f6ed8bc9128 in bin_bus_handler (bus=<optimized out>, message=<optimized out>, bin=<optimized out>) at ../subprojects/gstreamer/gst/gstbin.c:3286
   #21 0x00007f6ed8bdbae2 in gst_bus_post (bus=bus@entry=0xfdf500 [GstBus], message=message@entry=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstbus.c:359
   #22 0x00007f6ed8bf1396 in gst_element_post_message_default (element=element@entry=0xd705e0 [NleSource], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2067
   #23 0x00007f6ed8bccbee in gst_bin_post_message (element=0xd705e0 [NleSource], msg=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstmessage.h:376
   #24 0x00007f6ed8bf4b66 in gst_element_post_message (element=0xd705e0 [NleSource], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2110
   #25 0x00007f6ed8bc9128 in bin_bus_handler (bus=<optimized out>, message=<optimized out>, bin=<optimized out>) at ../subprojects/gstreamer/gst/gstbin.c:3286
   #26 0x00007f6ed8bdbae2 in gst_bus_post (bus=bus@entry=0x1042400 [GstBus], message=message@entry=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstbus.c:359
   #27 0x00007f6ed8bf1396 in gst_element_post_message_default (element=element@entry=0x1029450 [GstBin], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2067
   #28 0x00007f6ed8bccbee in gst_bin_post_message (element=0x1029450 [GstBin], msg=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstmessage.h:376
   #29 0x00007f6ed8bf4b66 in gst_element_post_message (element=0x1029450 [GstBin], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2110
   #30 0x00007f6ed8bc9128 in bin_bus_handler (bus=<optimized out>, message=<optimized out>, bin=<optimized out>) at ../subprojects/gstreamer/gst/gstbin.c:3286
   #31 0x00007f6ed8bdbae2 in gst_bus_post (bus=bus@entry=0x1042640 [GstBus], message=message@entry=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstbus.c:359
   #32 0x00007f6ed8bf1396 in gst_element_post_message_default (element=element@entry=0x7f6eb42fc7a0 [GstURIDecodeBin], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2067
   #33 0x00007f6ed8bccbee in gst_bin_post_message (element=0x7f6eb42fc7a0 [GstURIDecodeBin], msg=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstmessage.h:376
   #34 0x00007f6ed8bf4b66 in gst_element_post_message (element=0x7f6eb42fc7a0 [GstURIDecodeBin], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2110
   #35 0x00007f6ed8bc9128 in bin_bus_handler (bus=<optimized out>, message=<optimized out>, bin=<optimized out>) at ../subprojects/gstreamer/gst/gstbin.c:3286
   #36 0x00007f6ed8bdbae2 in gst_bus_post (bus=bus@entry=0x7f6eb80a7130 [GstBus], message=message@entry=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstbus.c:359
   #37 0x00007f6ed8bf1396 in gst_element_post_message_default (element=element@entry=0x7f6e6c02aa60 [GstDecodeBin], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2067
   #38 0x00007f6ed8bccbee in gst_bin_post_message (element=0x7f6e6c02aa60 [GstDecodeBin], msg=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstmessage.h:376
   #39 0x00007f6ed8bf4b66 in gst_element_post_message (element=0x7f6e6c02aa60 [GstDecodeBin], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2110
   #40 0x00007f6ec8f1e00d in gst_decode_bin_handle_message (bin=0x7f6e6c02aa60 [GstDecodeBin], msg=<optimized out>) at ../subprojects/gst-plugins-base/gst/playback/gstdecodebin2.c:5667
   #41 0x00007f6ed8bc9128 in bin_bus_handler (bus=<optimized out>, message=<optimized out>, bin=<optimized out>) at ../subprojects/gstreamer/gst/gstbin.c:3286
   #42 0x00007f6ed8bdbae2 in gst_bus_post (bus=bus@entry=0x7f6eb4139110 [GstBus], message=message@entry=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstbus.c:359
   #43 0x00007f6ed8bf1396 in gst_element_post_message_default (element=element@entry=0x7f6e54038c70 [GESDemux], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2067
   #44 0x00007f6ed8bccbee in gst_bin_post_message (element=0x7f6e54038c70 [GESDemux], msg=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstmessage.h:376
   #45 0x00007f6ed8bf4b66 in gst_element_post_message (element=0x7f6e54038c70 [GESDemux], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2110
   #46 0x00007f6ed8bc9128 in bin_bus_handler (bus=<optimized out>, message=<optimized out>, bin=<optimized out>) at ../subprojects/gstreamer/gst/gstbin.c:3286
   #47 0x00007f6ed8bdbae2 in gst_bus_post (bus=bus@entry=0x7f6eb4139350 [GstBus], message=message@entry=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstbus.c:359
   #48 0x00007f6ed8bf1396 in gst_element_post_message_default (element=element@entry=0x7f6e6c01ae50 [GESTimeline], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2067
   #49 0x00007f6ed8bccbee in gst_bin_post_message (element=0x7f6e6c01ae50 [GESTimeline], msg=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstmessage.h:376
   #50 0x00007f6ed8bf4b66 in gst_element_post_message (element=element@entry=0x7f6e6c01ae50 [GESTimeline], message=0x7f6eb411ac90) at ../subprojects/gstreamer/gst/gstelement.c:2110
   #51 0x00007f6ed8cfa221 in ges_timeline_change_state (element=0x7f6e6c01ae50 [GESTimeline], transition=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gst-editing-services/ges/ges-timeline.c:450
   #52 0x00007f6ed8bf5d1e in gst_element_change_state (element=element@entry=0x7f6e6c01ae50 [GESTimeline], transition=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:3033
   #53 0x00007f6ed8bf6868 in gst_element_continue_state (element=element@entry=0x7f6e6c01ae50 [GESTimeline], ret=ret@entry=GST_STATE_CHANGE_SUCCESS) at ../subprojects/gstreamer/gst/gstelement.c:2741
   #54 0x00007f6ed8bf5d67 in gst_element_change_state (element=element@entry=0x7f6e6c01ae50 [GESTimeline], transition=transition@entry=GST_STATE_CHANGE_NULL_TO_READY) at ../subprojects/gstreamer/gst/gstelement.c:3072
   #55 0x00007f6ed8bf6368 in gst_element_set_state_func (element=0x7f6e6c01ae50 [GESTimeline], state=GST_STATE_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:2987
   #56 0x00007f6ed8bd2129 in gst_bin_element_set_state (next=GST_STATE_PAUSED, current=GST_STATE_PAUSED, start_time=0, base_time=0, element=0x7f6e6c01ae50 [GESTimeline], bin=0x7f6e54038c70 [GESDemux]) at ../subprojects/gstreamer/gst/gstbin.c:2615
   #57 gst_bin_change_state_func (element=0x7f6e54038c70 [GESDemux], transition=GST_STATE_CHANGE_PAUSED_TO_PAUSED) at ../subprojects/gstreamer/gst/gstbin.c:2957
   #58 0x00007f6ed8bf5d1e in gst_element_change_state (element=element@entry=0x7f6e54038c70 [GESDemux], transition=transition@entry=GST_STATE_CHANGE_PAUSED_TO_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:3033
   #59 0x00007f6ed8bf6368 in gst_element_set_state_func (element=0x7f6e54038c70 [GESDemux], state=GST_STATE_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:2987
   #60 0x00007f6ed8bd2129 in gst_bin_element_set_state (next=GST_STATE_PAUSED, current=GST_STATE_READY, start_time=0, base_time=0, element=0x7f6e54038c70 [GESDemux], bin=0x7f6e6c02aa60 [GstDecodeBin]) at ../subprojects/gstreamer/gst/gstbin.c:2615
   #61 gst_bin_change_state_func (element=0x7f6e6c02aa60 [GstDecodeBin], transition=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gstreamer/gst/gstbin.c:2957
   #62 0x00007f6ec8f1e84f in gst_decode_bin_change_state (element=0x7f6e6c02aa60 [GstDecodeBin], transition=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gst-plugins-base/gst/playback/gstdecodebin2.c:5482
   #63 0x00007f6ed8bf5d1e in gst_element_change_state (element=element@entry=0x7f6e6c02aa60 [GstDecodeBin], transition=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:3033
   #64 0x00007f6ed8bf6868 in gst_element_continue_state (element=element@entry=0x7f6e6c02aa60 [GstDecodeBin], ret=ret@entry=GST_STATE_CHANGE_SUCCESS) at ../subprojects/gstreamer/gst/gstelement.c:2741
   #65 0x00007f6ed8bf5d67 in gst_element_change_state (element=element@entry=0x7f6e6c02aa60 [GstDecodeBin], transition=transition@entry=GST_STATE_CHANGE_NULL_TO_READY) at ../subprojects/gstreamer/gst/gstelement.c:3072
   #66 0x00007f6ed8bf6368 in gst_element_set_state_func (element=0x7f6e6c02aa60 [GstDecodeBin], state=GST_STATE_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:2987
   #67 0x00007f6ed8bf5ae8 in gst_element_sync_state_with_parent (element=0x7f6e6c02aa60 [GstDecodeBin]) at ../subprojects/gstreamer/gst/gstelement.c:2413
   #68 0x00007f6ed89f17a0 in g_slist_foreach (list=<optimized out>, func=0x7f6ed8bf5a50 <gst_element_sync_state_with_parent>, user_data=user_data@entry=0x0) at ../glib/gslist.c:880
   #69 0x00007f6ec8f37d45 in gst_uri_decode_bin_change_state (element=<optimized out>, transition=<optimized out>) at ../subprojects/gst-plugins-base/gst/playback/gsturidecodebin.c:2869
   #70 0x00007f6ed8bf5d1e in gst_element_change_state (element=element@entry=0x7f6eb42fc7a0 [GstURIDecodeBin], transition=transition@entry=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:3033
   #71 0x00007f6ed8bf6368 in gst_element_set_state_func (element=0x7f6eb42fc7a0 [GstURIDecodeBin], state=GST_STATE_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:2987
   #72 0x00007f6ed8bd2129 in gst_bin_element_set_state (next=GST_STATE_PAUSED, current=GST_STATE_READY, start_time=0, base_time=0, element=0x7f6eb42fc7a0 [GstURIDecodeBin], bin=0x1029450 [GstBin]) at ../subprojects/gstreamer/gst/gstbin.c:2615
   #73 gst_bin_change_state_func (element=0x1029450 [GstBin], transition=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gstreamer/gst/gstbin.c:2957
   #74 0x00007f6ed8bf5d1e in gst_element_change_state (element=element@entry=0x1029450 [GstBin], transition=transition@entry=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:3033
   #75 0x00007f6ed8bf6368 in gst_element_set_state_func (element=0x1029450 [GstBin], state=GST_STATE_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:2987
   #76 0x00007f6ed8bd2129 in gst_bin_element_set_state (next=GST_STATE_PAUSED, current=GST_STATE_READY, start_time=0, base_time=0, element=0x1029450 [GstBin], bin=0xd705e0 [NleSource]) at ../subprojects/gstreamer/gst/gstbin.c:2615
   #77 gst_bin_change_state_func (element=0xd705e0 [NleSource], transition=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gstreamer/gst/gstbin.c:2957
   #78 0x00007f6ec805533f in nle_object_change_state (element=0xd705e0 [NleSource], transition=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gst-editing-services/plugins/nle/nleobject.c:748
   #79 0x00007f6ed8bf5d1e in gst_element_change_state (element=element@entry=0xd705e0 [NleSource], transition=transition@entry=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:3033
   #80 0x00007f6ed8bf6368 in gst_element_set_state_func (element=0xd705e0 [NleSource], state=GST_STATE_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:2987
   #81 0x00007f6ed8bd2129 in gst_bin_element_set_state (next=GST_STATE_PAUSED, current=GST_STATE_READY, start_time=0, base_time=0, element=0xd705e0 [NleSource], bin=0x1029110 [GstBin]) at ../subprojects/gstreamer/gst/gstbin.c:2615
   #82 gst_bin_change_state_func (element=0x1029110 [GstBin], transition=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gstreamer/gst/gstbin.c:2957
   #83 0x00007f6ed8bf5d1e in gst_element_change_state (element=element@entry=0x1029110 [GstBin], transition=transition@entry=GST_STATE_CHANGE_READY_TO_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:3033
   #84 0x00007f6ed8bf6368 in gst_element_set_state_func (element=0x1029110 [GstBin], state=GST_STATE_PAUSED) at ../subprojects/gstreamer/gst/gstelement.c:2987
   #85 0x00007f6ed8bf5ae8 in gst_element_sync_state_with_parent (element=0x1029110 [GstBin]) at ../subprojects/gstreamer/gst/gstelement.c:2413
   #86 0x00007f6ec8060356 in _activate_new_stack (toplevel_seek=<optimized out>, comp=0x10261d0 [NleComposition]) at ../subprojects/gst-editing-services/plugins/nle/nlecomposition.c:3117
   #87 update_pipeline (comp=comp@entry=0x10261d0 [NleComposition], currenttime=<optimized out>, seqnum=<optimized out>, update_reason=update_reason@entry=COMP_UPDATE_STACK_INITIALIZE) at ../subprojects/gst-editing-services/plugins/nle/nlecomposition.c:3396
   #88 0x00007f6ec80614f6 in _initialize_stack_func (comp=0x10261d0 [NleComposition], ucompo=0x108c800) at ../subprojects/gst-editing-services/plugins/nle/nlecomposition.c:732
   #89 0x00007f6ed893788a in g_closure_invoke (closure=<optimized out>, return_value=<optimized out>, n_param_values=<optimized out>, param_values=<optimized out>, invocation_hint=<optimized out>) at ../gobject/gclosure.c:810
   #90 0x00007f6ec805aaf6 in _execute_actions (comp=0x10261d0 [NleComposition]) at ../subprojects/gst-editing-services/plugins/nle/nlecomposition.c:412
   #91 0x00007f6ed8c4c1cf in gst_task_func (task=0x7f6e6c01c290 [GstTask]) at ../subprojects/gstreamer/gst/gsttask.c:328
   #92 0x00007f6ed89fc0f4 in g_thread_pool_thread_proxy (data=<optimized out>) at ../glib/gthreadpool.c:354
   #93 0x00007f6ed89fb7f2 in g_thread_proxy (data=0x7f6eb0017800) at ../glib/gthread.c:807
   #94 0x00007f6ed7e14432 in start_thread () at /lib64/libpthread.so.0
   #95 0x00007f6ed86779d3 in clone () at /lib64/libc.so.6
```

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/167>

4 years agocheck: give nle_tempochange test more time
Henry Wilkes [Tue, 14 Apr 2020 09:22:09 +0000 (10:22 +0100)]
check: give nle_tempochange test more time

These test can take longer than most under valgrind, so give them a
little more time until they timeout.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>

4 years agotimeline: fix adding track when layers contains clips
Henry Wilkes [Mon, 13 Apr 2020 10:40:55 +0000 (11:40 +0100)]
timeline: fix adding track when layers contains clips

Made sure that adding a new track only uses select-tracks-for-object for
core children to determine whether a track elements should be added to the
new track or not, and *not* any other track. In particular, there should
be *no* change in the existing tracks of the timeline when adding another
track. Moreover, a new track should not invoke the creation of track
elements for other tracks.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>

4 years agonleobject: stop using media-duration-factor
Henry Wilkes [Wed, 8 Apr 2020 16:11:14 +0000 (17:11 +0100)]
nleobject: stop using media-duration-factor

The property had been deprecated and is unused.

This property is not needed. Any internal time effect that an nleoperation
wraps is itself responsible for converting seek/segment timestamps.
Previously, the ghostpads were performing a rate conversion after the
rate element had already done so, essentially doubling their effect on
seeks and segment times. This was always unnecessary, but went unnoticed
by the tempochange test because it was using an identity element rather
than an actual rate-changing element.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>

4 years agonleoperation: stop setting next_base_time
Henry Wilkes [Wed, 8 Apr 2020 16:08:41 +0000 (17:08 +0100)]
nleoperation: stop setting next_base_time

This property was unused.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>

4 years agonlecomposition: Fix seeking with stop
Thibault Saunier [Tue, 21 Apr 2020 20:22:31 +0000 (16:22 -0400)]
nlecomposition: Fix seeking with stop

And add some tests

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/166>

4 years agolaunch: Add support for testfiles
Thibault Saunier [Fri, 24 Apr 2020 21:15:16 +0000 (17:15 -0400)]
launch: Add support for testfiles

Making it simpler to define a test in a single files, including the
configuration etc..

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/166>

4 years agotrack: Do not commit ourselves automatically when changing state from wrong thread
Thibault Saunier [Fri, 24 Apr 2020 20:46:50 +0000 (16:46 -0400)]
track: Do not commit ourselves automatically when changing state from wrong thread

The user is responsible to commit the timeline from the right thread
in that case and in the case of gesdemux, the loaded timeline is filling
gaps automatically when the project is set loaded.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/166>

4 years agomacos: fix python's configure checks
Andoni Morales Alastruey [Sat, 18 Apr 2020 14:22:25 +0000 (16:22 +0200)]
macos: fix python's configure checks

4 years agoges: Fix interlaced stream playback
Thibault Saunier [Fri, 17 Apr 2020 16:35:26 +0000 (12:35 -0400)]
ges: Fix interlaced stream playback

Negotiation was failling as `videoflip` was not allowing not
progressive interlacing.

Also avoid adding a deinterlace element when it is useless.

4 years agoframepositioner: Fix some source repositionning rounding issues
Thibault Saunier [Fri, 17 Apr 2020 00:27:30 +0000 (20:27 -0400)]
framepositioner: Fix some source repositionning rounding issues

Avoid loosing (too much) precision when rescaling back and forth by
storing values in gdoubles.

Handle the fact that position values can be negative

Also fix debug category static variable
as it clashes with the instance variable name in a few methods.

4 years agotimeline: Fix wrong usage of scale_int
Thibault Saunier [Thu, 16 Apr 2020 16:53:00 +0000 (12:53 -0400)]
timeline: Fix wrong usage of scale_int

We are multiplying the framerate by GST_SECOND and thus have no
guarantee that it won't overflow.

4 years agoges: Place imagefreeze at right place
Thibault Saunier [Sat, 11 Apr 2020 15:40:06 +0000 (11:40 -0400)]
ges: Place imagefreeze at right place

Negotiation fails when having the imagefreeze after videorate and
frame positioning won't happen after seeks if we do not put it
before the postioner

4 years agoasset: Avoid dereferencing NULL pointer
Thibault Saunier [Thu, 9 Apr 2020 15:24:44 +0000 (11:24 -0400)]
asset: Avoid dereferencing NULL pointer

CID 1461286

4 years agobasebin: Do not set stream_group if upstream didn't provide it
Thibault Saunier [Thu, 9 Apr 2020 15:20:34 +0000 (11:20 -0400)]
basebin: Do not set stream_group if upstream didn't provide it

CID: 1461278

4 years agoframepositionner: Fix wrong old size check condition
Thibault Saunier [Thu, 9 Apr 2020 15:17:59 +0000 (11:17 -0400)]
framepositionner: Fix wrong old size check condition

CID: 1461277

4 years agovalidate: Avoid dereferencing NULL pointer
Thibault Saunier [Thu, 9 Apr 2020 15:16:34 +0000 (11:16 -0400)]
validate: Avoid dereferencing NULL pointer

CID: 1461266

4 years agotimeline: Ensure setting framerate in timeline_get_framerate
Thibault Saunier [Thu, 9 Apr 2020 15:10:43 +0000 (11:10 -0400)]
timeline: Ensure setting framerate in timeline_get_framerate

CID: 14612501461288

4 years agotests: Check that linking pads works
Thibault Saunier [Thu, 9 Apr 2020 15:07:04 +0000 (11:07 -0400)]
tests: Check that linking pads works

CID: 1456061

4 years agostructured-interface: Properly check that setting keyframe works
Thibault Saunier [Thu, 9 Apr 2020 15:02:26 +0000 (11:02 -0400)]
structured-interface: Properly check that setting keyframe works

Fixes CID: 1455490

4 years agogesdemux: Check result of g_stat
Thibault Saunier [Thu, 9 Apr 2020 14:59:40 +0000 (10:59 -0400)]
gesdemux: Check result of g_stat

CID: 14554891455521

4 years agotest: tempochange: Plug leak
Thibault Saunier [Thu, 9 Apr 2020 14:54:26 +0000 (10:54 -0400)]
test: tempochange: Plug leak

CID: 1455448

4 years agoges: Cast to signed int to compare agasint 0
Thibault Saunier [Thu, 9 Apr 2020 14:42:03 +0000 (10:42 -0400)]
ges: Cast to signed int to compare agasint 0

The check made sense but we were not casting to be able to check
signess of subtraction result.

CID: 1444923

4 years agotimeline: Do not compare unsigned with 0
Thibault Saunier [Thu, 9 Apr 2020 14:37:20 +0000 (10:37 -0400)]
timeline: Do not compare unsigned with 0

Layer priorities are always positive the check was making no
sense in any case.

Fixes CID: 14449221461284

4 years agoges: Always check return value of `ges_container_add`
Thibault Saunier [Thu, 9 Apr 2020 14:31:36 +0000 (10:31 -0400)]
ges: Always check return value of `ges_container_add`

Making coverity happy

CIDs: 14614601461461146146214614631461464146146514614661461468,

4 years agoges: Fix sending EOS on testclip when using timeoverlay
Thibault Saunier [Thu, 9 Apr 2020 14:00:43 +0000 (10:00 -0400)]
ges: Fix sending EOS on testclip when using timeoverlay

Basically when using timeoverlay we where waiting for input-selector
to receive EOS on its active on the output-selector streaming thread
but... EOS was being sent from that same thread waiting for input-selector
to unblock to send EOS on its other pad.

In our specific use case we want EOS to be sent only on the active pad.

Fixes: https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/issues/103
4 years agoges: Update documentation
Thibault Saunier [Thu, 9 Apr 2020 13:29:17 +0000 (09:29 -0400)]
ges: Update documentation

And start generating TrackElement children property with a stupid
simple script

4 years agoges: Add the foreground color child property
Thibault Saunier [Thu, 9 Apr 2020 13:24:12 +0000 (09:24 -0400)]
ges: Add the foreground color child property

4 years agoges: Fix reloading UriClipAsset synchronously
Thibault Saunier [Tue, 7 Apr 2020 14:53:15 +0000 (10:53 -0400)]
ges: Fix reloading UriClipAsset synchronously

And add tests for that

4 years agoges: Rework the way we ensure core elements are not wrongly moved between clips
Thibault Saunier [Tue, 7 Apr 2020 14:47:07 +0000 (10:47 -0400)]
ges: Rework the way we ensure core elements are not wrongly moved between clips

Instead of focusing on the instances of the clips and their children,
we relax the check to allow moving track element clip between clips
that share a common asset. This makes it as correct conceptually but
more flexible, and the code becomes simpler.

4 years agogroup: tidied timeline membership in copy-paste
Henry Wilkes [Thu, 2 Apr 2020 10:58:18 +0000 (11:58 +0100)]
group: tidied timeline membership in copy-paste

Previously, the GESContainer ->paste method and GESGroup ->paste methods
were unnecessarily setting the timeline of groups, even though this is
handled by the GESGroup ->child_added method. This could result in the
group being added multiple times.

4 years agotimeline-tree: fix overlap check
Henry Wilkes [Wed, 1 Apr 2020 20:34:48 +0000 (21:34 +0100)]
timeline-tree: fix overlap check

Previously, the code was not able to detect that an element overlaps on
its end, nor could it detect that an element overlaps two elements that
already overlap.

4 years agoclip: tidy grouping
Henry Wilkes [Mon, 6 Apr 2020 11:44:30 +0000 (12:44 +0100)]
clip: tidy grouping

Make the grouping of clips cleaner by checking that the clips share the
same asset.

4 years agoclip: secure adding clip to layer
Henry Wilkes [Mon, 6 Apr 2020 11:42:03 +0000 (12:42 +0100)]
clip: secure adding clip to layer

Add more checks when adding a clip to a layer, or moving a clip to a new
layer. Also, mark the "layer" property as explicit-notify.

4 years agouri-clip: match children by track
Henry Wilkes [Mon, 6 Apr 2020 11:28:13 +0000 (12:28 +0100)]
uri-clip: match children by track

When the asset of a uri clip is reset, its core children are removed and
replaced by the new core children. When replacing, the `set_asset`
method attempts to copy children properties from the previous children
to the new children. However, the children were matched by track-type
only. This would not function as intended when a URI contains multiple
audio or video streams. Instead, we now match children by the tracks
themselves. This should work better, provided the user's
select-tracks-for-object is well behaved.

Also, fix a memory problem in `set_mute` for when a child is not in a
track.

4 years agotimeline-element: only copy read-write properties
Henry Wilkes [Mon, 6 Apr 2020 11:26:11 +0000 (12:26 +0100)]
timeline-element: only copy read-write properties

Only copy the properties that can be both read and written, and are not
construct only. Similarly for child properties when a track-element is
deep copied.

4 years agotimeline: return sunk element on pasting
Henry Wilkes [Mon, 6 Apr 2020 11:17:43 +0000 (12:17 +0100)]
timeline: return sunk element on pasting

Technically, an element can still be floating on the return from
`->paste` (e.g. a clip not in a layer). Since the return of the `_paste`
methods are (return full) a non-floating object is probably expected in
all cases.

4 years agoauto-transition: select track directly
Henry Wilkes [Mon, 6 Apr 2020 11:16:11 +0000 (12:16 +0100)]
auto-transition: select track directly

By-pass the select-tracks-for-object signal for auto-transitions since
their track element must land in the same track as the elements it is
the auto-transition for.

4 years agotimeline: re-handle clip children track selection
Henry Wilkes [Mon, 6 Apr 2020 11:09:54 +0000 (12:09 +0100)]
timeline: re-handle clip children track selection

The way a clip's track elements are added to tracks was re-handled. This
doesn't affect the normal usage of a simple audio-video timeline, where
the tracks are added before any clips, but usage for multi-track
timelines has improved. The main changes are:

+ We can now handle a track being selected for more than one track,
  including a full copy of their children properties and bindings.
  (Previously broken.)
+ When a clip is split, we copy the new elements directly into the same
  track, avoiding select-tracks-for-object.
+ When a clip is grouped or ungrouped, we avoid moving the elements to
  or from tracks.
+ Added API to allow users to copy the core elements of a clip directly
  into a track, complementing select-tracks-for-object.
+ Enforced the rule that a clip can only contain one core child in a
  track, and all the non-core children must be added to tracks that
  already contains a core child. This extends the previous condition
  that two sources from the same clip should not be added to the same
  track.
+ Made ges_track_add_element check that the newly added track element
  does not break the configuration rules of the timeline.
+ When adding a track to a timeline, we only use
  select-tracks-for-object to check whether track elements should be
  added to the new track, not existing ones.
+ When removing a track from a timeline, we empty it of all the track
  elements that are controlled by a clip. Thus, we ensure that a clip
  only contains elements that are in the tracks of the same timeline, or
  no track. Similarly, when removing a clip from a timeline.
+ We can now avoid unsupported timeline configurations when a layer is
  added to a timeline, and already contains clips.
+ We can now avoid unsupported timeline configurations when a track is
  added to a timeline, and the timeline already contains clips.

Fixes https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/issues/84

4 years agotimeline: stop connecting to track-element-added
Henry Wilkes [Thu, 26 Mar 2020 09:21:42 +0000 (09:21 +0000)]
timeline: stop connecting to track-element-added

This was used to connect to the track element's notify::start signal in
order to update the duration of the timeline (it is not clear why the
notify::duration signal was not also connected to for the same reason).
However, this is already covered by the timeline_tree_move method, which
is always called to update the start of a track element, even if it is not
part of a clip (and similarly for timeline_tree_trim, which is called
when the duration is set).

4 years agoclip: allow arbitrary max-duration when no core children
Henry Wilkes [Wed, 25 Mar 2020 19:35:11 +0000 (19:35 +0000)]
clip: allow arbitrary max-duration when no core children

Before the max-duration could be set arbitrarily when the clip was empty,
to indicate what the max-duration would be once the core children were
created. Now, we can also do this whilst the clip only contains non-core
children.

4 years agotrack-element: change owner to creator
Henry Wilkes [Wed, 25 Mar 2020 18:49:16 +0000 (18:49 +0000)]
track-element: change owner to creator

Rename the private "owners" to "creators" to avoid confusing this with
the owner of the track element's memory.

Also made the ungroup method for GESClip symmetric by making all the
children of the resulting clips share their creators, which allows them
to be added to any of the other ungrouped clips. Once the clips are
grouped back together, the tracks loose these extra creators.

4 years agocontainer: change ownership when adding
Henry Wilkes [Mon, 6 Apr 2020 11:21:54 +0000 (12:21 +0100)]
container: change ownership when adding

Make sure we sink the child on adding, and keep it alive until the end
in case the method fails.

Also, since the child mappings hold a ref to the child, they should give
them up in their free method. This way, the ref will be given up on
disposing, even if ges_container_remove fails.

Also, reverse setting of the start of the container if adding fails.

4 years agoges: Fix build with GCC 10
Nirbheek Chauhan [Mon, 6 Apr 2020 17:36:29 +0000 (23:06 +0530)]
ges: Fix build with GCC 10

gcc-10 defaults to -fno-common, which exposes a symbol conflict, so
use `static` correctly. Also we don't use `parent_extractable_iface`
in `ges-uri-clip.c`.
See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85678

4 years agoges: Fix trimming clip inside deeply nested groups
Thibault Saunier [Tue, 31 Mar 2020 14:25:49 +0000 (11:25 -0300)]
ges: Fix trimming clip inside deeply nested groups

This broke in 6b7c658b6a551a5b9170987ba44592d1d819e1ae

4 years agouri-clip: Remove dead code
Thibault Saunier [Wed, 25 Mar 2020 01:47:01 +0000 (22:47 -0300)]
uri-clip: Remove dead code

GES_TESTING_ASSETS_DIRECTORY is prehistoric and since then
new mechanism for asset relocation have been added, it makes
no sense to keep that unused code path

4 years agouri-clip: Remove ->create_track_element implementation
Thibault Saunier [Wed, 25 Mar 2020 01:44:07 +0000 (22:44 -0300)]
uri-clip: Remove ->create_track_element implementation

It is dead code

4 years agoges: Deprecate GESImageSource and GESMultiFileSource
Thibault Saunier [Wed, 25 Mar 2020 01:35:35 +0000 (22:35 -0300)]
ges: Deprecate GESImageSource and GESMultiFileSource

Refactoring GESVideoSource so that #GESUriVideoSource can handle
still image in a simple way

MultiFileSource has been replaced with the new `imagesequencesrc`
element, this was totally broken anyway as `multifilesrc` can not seek
properly.

4 years agotrack-element: Create nleobject on GESExtractable::set_asset
Thibault Saunier [Wed, 25 Mar 2020 01:30:38 +0000 (22:30 -0300)]
track-element: Create nleobject on GESExtractable::set_asset

This means that we have all the information about the asset
when constructing the underlying GstElements.

This also allows to cleanup some code all around

4 years agotimeline:element: Refactor the way we 'copy'
Thibault Saunier [Wed, 25 Mar 2020 01:25:47 +0000 (22:25 -0300)]
timeline:element: Refactor the way we 'copy'

Simplifying the implementation and making sure assets are set asap

4 years agoges: Use assets to instantiate track elements/group
Thibault Saunier [Wed, 25 Mar 2020 01:23:16 +0000 (22:23 -0300)]
ges: Use assets to instantiate track elements/group

And deprecate all GESTrackElement constructors, but the GESEffect one.
Those should **never** be created by users and should become internal
in the future.

Stop having docstring for the constructors that were internal.

4 years agotests: Cleanup test files handling
Thibault Saunier [Wed, 18 Mar 2020 19:24:08 +0000 (16:24 -0300)]
tests: Cleanup test files handling

4 years agoformatter: Serialize source properties
Thibault Saunier [Fri, 13 Mar 2020 18:03:17 +0000 (15:03 -0300)]
formatter: Serialize source properties

This way we ensure that the TrackElement 'active' property is
properly serialized

4 years agoges: Add a way to set layer activeness by track
Thibault Saunier [Fri, 6 Mar 2020 21:56:52 +0000 (18:56 -0300)]
ges: Add a way to set layer activeness by track

a.k.a muting layers.

Adding unit tests and making sure serialization works properly

4 years agoelement: Add API safe guard against invalid position in edit()
Thibault Saunier [Tue, 24 Mar 2020 00:21:10 +0000 (21:21 -0300)]
element: Add API safe guard against invalid position in edit()

4 years agovalidate: Refactor actions implementation
Thibault Saunier [Tue, 24 Mar 2020 00:11:45 +0000 (21:11 -0300)]
validate: Refactor actions implementation

Making them simpler to read and avoiding leaks

4 years agostructured-interface: Fix adding clip to layer error reporting
Thibault Saunier [Mon, 23 Mar 2020 18:14:13 +0000 (15:14 -0300)]
structured-interface: Fix adding clip to layer error reporting

4 years agoges: Add a SourceClipAsset class
Thibault Saunier [Tue, 17 Mar 2020 14:53:47 +0000 (11:53 -0300)]
ges: Add a SourceClipAsset class

Cleaning up the way we use the default framerate for natural
frame rate.

4 years agolaunch: Add a way to disable validate at runtime
Thibault Saunier [Tue, 10 Mar 2020 19:10:12 +0000 (16:10 -0300)]
launch: Add a way to disable validate at runtime

Also avoid to add useless bin in our sinks

4 years agoges: Plug some leaks
Thibault Saunier [Mon, 9 Mar 2020 18:38:58 +0000 (15:38 -0300)]
ges: Plug some leaks

4 years agovalidate: Add support to seek in frames
Thibault Saunier [Fri, 28 Feb 2020 14:56:22 +0000 (11:56 -0300)]
validate: Add support to seek in frames

4 years agoges: support test clips assets natural size/framerate
Thibault Saunier [Fri, 28 Feb 2020 14:47:25 +0000 (11:47 -0300)]
ges: support test clips assets natural size/framerate

This way we can test this kind of behaviour without requiring
real sources.

Also add simple tests.

4 years agoges: Add APIs to have a sens of frame numbers
Thibault Saunier [Fri, 21 Feb 2020 12:17:11 +0000 (09:17 -0300)]
ges: Add APIs to have a sens of frame numbers

APIs:
   - ges_timeline_get_frame_time
   - ges_timeline_get_frame_at
   - ges_clip_asset_get_frame_time
   - ges_clip_get_timeline_time_from_source_frame

Extracting ges_util_structure_get_clocktime to internal utilities adding
support for specifying timing values in frames with the special
f<frame-number> synthax.

4 years agoutils: fix argument sanitization
Henry Wilkes [Tue, 29 Oct 2019 16:52:52 +0000 (16:52 +0000)]
utils: fix argument sanitization

_sanitize_argument is supposed to wrap arguments in '"' quote marks such
that they can be parsed and copied into a GstStructure string. This
purpose is now supported more directly, which fixes some bugs, e.g.:
  arguments                before                  fix
+title my=title          +title my="title"       +title "my=title"
+title abc n=my=name     +title abc n="my="name" +title abc n="my=name"
+title my"title          +title "my"title"       +title "my\"title"
+title my\title          +title "my\title"       +title "my\\title"

4 years agolaunch: Fix memory management issue with the rendering format
Thibault Saunier [Fri, 28 Feb 2020 14:52:38 +0000 (11:52 -0300)]
launch: Fix memory management issue with the rendering format

4 years agovalidate: Rename edit-container to edit
Thibault Saunier [Tue, 25 Feb 2020 20:42:47 +0000 (17:42 -0300)]
validate: Rename edit-container to edit

Keeping the old version for backward compat

4 years agoges: Add a timeoverlay to video test sources
Thibault Saunier [Fri, 21 Feb 2020 20:17:10 +0000 (17:17 -0300)]
ges: Add a timeoverlay to video test sources

This is often very useful to have a timeoverlay inside test sources.
We do not want to use it as an effect as segments are not the sames
in GES when it comes to nleoperations.

4 years agoelement: Handle using own property as child property
Thibault Saunier [Tue, 25 Feb 2020 21:39:47 +0000 (18:39 -0300)]
element: Handle using own property as child property

Avoiding ref cycles

4 years agoges: Ensure GESClips assets are always ClipAssets
Thibault Saunier [Fri, 21 Feb 2020 20:16:01 +0000 (17:16 -0300)]
ges: Ensure GESClips assets are always ClipAssets

4 years agoges: Add API to retrieve the natural framerate of an element
Thibault Saunier [Tue, 18 Feb 2020 18:21:38 +0000 (15:21 -0300)]
ges: Add API to retrieve the natural framerate of an element

4 years agoges: Some memory management fixes setting track mixing
Thibault Saunier [Fri, 28 Feb 2020 20:53:55 +0000 (17:53 -0300)]
ges: Some memory management fixes setting track mixing

Also fix 'mixing' property notifies

4 years agoges: Cleanup GESEdge and GESEditMode GEnum values
Thibault Saunier [Fri, 28 Feb 2020 20:50:05 +0000 (17:50 -0300)]
ges: Cleanup GESEdge and GESEditMode GEnum values

By duplicating the registered values, so that bindings have
better values to use

4 years agolaunch: Make command line provided sinks override scenario defined ones
Thibault Saunier [Mon, 2 Mar 2020 17:35:33 +0000 (14:35 -0300)]
launch: Make command line provided sinks override scenario defined ones

4 years agoframepositioner: Avoid dereferencing NULL pointer
Thibault Saunier [Fri, 28 Feb 2020 14:58:30 +0000 (11:58 -0300)]
framepositioner: Avoid dereferencing NULL pointer

4 years agovalidate: Initialize GValue before calling g_object_get_value
Thibault Saunier [Wed, 4 Mar 2020 19:03:30 +0000 (16:03 -0300)]
validate: Initialize GValue before calling g_object_get_value

This is required with GLib < 2.60

4 years agoges: Use #pragma once everywhere
Thibault Saunier [Tue, 17 Mar 2020 21:13:51 +0000 (18:13 -0300)]
ges: Use #pragma once everywhere

4 years agoges: Cleanup the way we declare object types
Thibault Saunier [Tue, 17 Mar 2020 18:51:39 +0000 (15:51 -0300)]
ges: Cleanup the way we declare object types

We create our own _DECLARE_ macro because we have instance structures

4 years agoges: Stop using hash_table_steal_extended
Thibault Saunier [Thu, 19 Mar 2020 12:15:07 +0000 (09:15 -0300)]
ges: Stop using hash_table_steal_extended

This appeard in GLib 2.58

Fixes https://gitlab.freedesktop.org/gstreamer/gst-editing-services/issues/99

4 years agoclip: Allow setting max-duration clips without TrackElements
Thibault Saunier [Wed, 18 Mar 2020 16:36:47 +0000 (13:36 -0300)]
clip: Allow setting max-duration clips without TrackElements

Otherwise this breaks quite a few assumption in user code, several
pitivi tests broke because of that.

4 years agoges: Make it so core elements can be re added to their 'owners'
Thibault Saunier [Wed, 18 Mar 2020 15:56:06 +0000 (12:56 -0300)]
ges: Make it so core elements can be re added to their 'owners'

The user might want to add/remove/add core children to clips and be able
to regroup ungrouped clip. This is needed for undo/redo in Pitivi for
example

4 years agocontainer: Let subclass know adding child was interrupted
Thibault Saunier [Wed, 18 Mar 2020 14:12:55 +0000 (11:12 -0300)]
container: Let subclass know adding child was interrupted

When the `child-added` signal emission was called, the
`GESContainer->child_added` vmethod was called (the signal is
`G_SIGNAL_RUN_FIRST`) so we need to call `GESContainer->child_removed`
ourself so subclasses know they do not control the child anymore.

4 years agotimeline-element: make start and duration EXPLICIT_NOTIFY
Henry Wilkes [Tue, 10 Mar 2020 16:01:02 +0000 (16:01 +0000)]
timeline-element: make start and duration EXPLICIT_NOTIFY

The properties will only have their signal emitted when they change in
value, even when g_object_set, etc, methods are used.

The _set_start method already did this, but start was missing the
EXPLICIT_NOTIFY flag. There should be no need to check that the property
has changed in ->set_start or ->set_duration

4 years agotimeline-element: make max-duration cap in-point
Henry Wilkes [Tue, 10 Mar 2020 15:27:20 +0000 (15:27 +0000)]
timeline-element: make max-duration cap in-point

Do not allow the in-point to exceed the max-duration of any timeline
element.

4 years agoclip: only allow children with the same timeline
Henry Wilkes [Tue, 10 Mar 2020 11:53:09 +0000 (11:53 +0000)]
clip: only allow children with the same timeline

Refuse the addition of children whose timeline is neither NULL nor the
clip's timeline.

4 years agoclip: re-handle child in-point and max-duration
Henry Wilkes [Tue, 10 Mar 2020 11:38:58 +0000 (11:38 +0000)]
clip: re-handle child in-point and max-duration

The in-point of a clip is kept in sync with its core children, unless they
have no has-internal-source.

The max-duration is defined as the minimum max-duration amongst the
clip's core children. If it is set to a new value, this sets the
max-duration of its core children to the same value if they have
has-internal-source set as TRUE.

Non-core children (such as effects on a source clip) do not influence
these values.

As part of this, we no longer track in-point in GESContainer. Unlike start
and duration, the in-point of a timeline element does not refer to its
extent in the timeline. As such, it has little meaning for most
collections of timeline-elements, in particular GESGroups. As such, there
is no generic way to relate the in-point of a container to its children.

4 years agotimeline-element: make in-point and max-duration EXPLICIT_NOTIFY
Henry Wilkes [Tue, 10 Mar 2020 11:35:23 +0000 (11:35 +0000)]
timeline-element: make in-point and max-duration EXPLICIT_NOTIFY

As such, they only emit a signal if their value changes, either through
their _set_inpoint or _set_max_duration methods, or through
g_object_set, etc.

Also, we now require the ->set_max_duration method to be implemented.
This was added to GESGroup, which will only allow the max-duration to be
set to GST_CLOCK_TIME_NONE.

4 years agotrack-element: add has-internal-source property
Henry Wilkes [Tue, 10 Mar 2020 11:29:40 +0000 (11:29 +0000)]
track-element: add has-internal-source property

Unless this property is set to TRUE, the in-point must be 0 and the
max-duration must be GST_CLOCK_TIME_NONE.

Also added EXPLICIT_NOTIFY flags to the active and track-type
properties such that their notifies are emitted only if the property
changes, even when the g_object_set, etc, methods are used.

Also added a missing notify signal to the set_active method.

4 years agoclip: copy and paste control bindings
Henry Wilkes [Tue, 3 Mar 2020 18:00:51 +0000 (18:00 +0000)]
clip: copy and paste control bindings

Previously the control bindings were not properly copied into the pasted
clip. Also changed the order so that elements are added to the clip
before the clip is added to the timeline.

4 years agotimeline-element: add signals for child properties
Henry Wilkes [Tue, 3 Mar 2020 14:31:10 +0000 (14:31 +0000)]
timeline-element: add signals for child properties

Add the child-property-added and child-property-removed signals to
GESTimelineElement.

GESContainer is able to use this to keep their child properties in sync
with their children: if they are added or removed from the child, they
are also added or removed from the container.

4 years agocontainer: freeze notifies during add and remove
Henry Wilkes [Mon, 2 Mar 2020 12:23:07 +0000 (12:23 +0000)]
container: freeze notifies during add and remove

Hold the notify signals for the container and the children until after
the child has been fully added or removed.

After the previous commit, this was used to ensure that the
notify::priority signal was sent for children of a clip *after* the
child-removed signal. This stopped being the case when the code in
->child_removed was moved to ->remove_child (the latter is called before
the child-removed signal is emitted, whilst the former is called
afterwards). Rather than undo this move of code, which was necessary to
ensure that ->add_child was always reversed, the notify::priority signal
is now simply delayed until after removing the child has completed. This
was done for all notify signals, as well as in the add method, to ensure
consistency.

This allows the test_clips.py test_signal_order_when_removing_effect to
pass.

Also make subclasses take a copy of the list of the children before
setting the start and duration, since this can potentially re-order the
children (if they have the SET_SIMPLE flag set).

4 years agoclip: make remove_child a reverse of add_child
Henry Wilkes [Mon, 2 Mar 2020 13:35:20 +0000 (13:35 +0000)]
clip: make remove_child a reverse of add_child

Previously, we relied on ->child_removed to reverse the priority changes
that occured in ->add_child. However, ->child_removed is not always
called (the signal child-removed is not always emitted) when a
->add_child needs to be removed. However, ->remove_child is always
called to reverse ->add_child, so the code was moved here. Otherwise, we
risk that the priorities of the clip will contain gaps, which will cause
problems when another child is added to the clip.